Originally Posted by
Brian Ratliff
See, I don't like statistical methods. it's a problem I have, particularly as an engineer. Statistics are a black box. Samples go in, lots of information, the black box does its thing and an answer spits out, throwing away 99.99% of that information to arrive at a couple numbers that look good on a chart. None of the outputs are based in the reality of physics and form. It's all just data manipulation; massaging data samples a certain way to uncover the parameters of an a priori defined pattern. Too often, people use that manipulated data in its raw form, without considering actual physical cause in making decisions. In my opinion, this is wrong. Using the data manipulation black box to guide an exploration of physics and form is fine; mistaking that manipulated data for real knowledge about the world is wrong.
I wonder if algebra, geometry, trigonometry, the calculus, etc. were regarded with so much animosity, suspicion, misunderstanding, and false reporting when they were first developed/introduced. Those who are knowledgeable about statistical design of experimentation and treatment of experimental results know the techniques are not at all a black box. It is perfectly possible to perform the necessary analyses and calculations by hand for simple cases so that the validity of the methods can be demonstrated to the researcher and acknowledged by them. Adherents to statistical methods also know that it is only by applying those methods that underlying physical causes can actually be uncovered and understood.
Far from using the statistical methods to validate an
a priori position, the usual result of statistical methods is to negate such prejudices. Nothing is more immune to experimenter prejudice than statistical methods of experimentation.